Surveillance sanitaire à partir de données des services d'urgence : modélisation de séries temporelles et analyse automatique

Abstract : Before the 2003 heat wave, French public health surveillance systems were mainly focused on known diseases (notifiable diseases, influenza, gastroenteritis...). Considering the important health consequences of this heat wave, the objective of health surveillance systems has become more general. A non-specific surveillance system has been developed to prevent unexpected health events. The goal is to detect any unusual sanitary event, in order to prevent its effects. The diversity and the large number of count data analysed daily in such system need to implement statistical methods routinely in order to trigger an alarm when the value of the observed count data deviates from its usual level. The objectives of this training course were to compare and evaluate the performance of three statistical methods for continuous monitoring of hospital activity data (Hutwagner's cusum control charts, Farrington's Algorithm, Arima method), and to establish a non-specific monitoring system based on the most efficient methods. A specific surveillance system based on the Serfling's regression method was also proposed for strengthen the existing control system. This comparison was made with semi-artificial time series of hospital activity count data. Several indicators, including sensitivity and specificity were calculated. The study results have showed a predominance of methods using large historical data (Farrington and Arima). However, given the wide variety of time series and surveillance objectives, it seems useful to analyse data using several methods.
Document type :
Master Thesis
Complete list of metadatas

Cited literature [16 references]  Display  Hide  Download

https://dumas.ccsd.cnrs.fr/dumas-00516268
Contributor : Ufr de Mathématique-Informatique / Irma <>
Submitted on : Thursday, September 9, 2010 - 9:48:59 AM
Last modification on : Thursday, December 14, 2017 - 9:56:52 AM
Long-term archiving on : Friday, December 10, 2010 - 2:41:01 AM

Identifiers

  • HAL Id : dumas-00516268, version 1

Collections

MS | DUMAS

Citation

Nicolas Jung. Surveillance sanitaire à partir de données des services d'urgence : modélisation de séries temporelles et analyse automatique. Méthodologie [stat.ME]. 2010. ⟨dumas-00516268⟩

Share

Metrics

Record views

315

Files downloads

484